Lauren Dowdeswell and Lucianna Churchill, University Centre Reaseheath
Dog harnesses are a popular walking aid for pet dogs. Research has focused on the biomechanical impact of assistance dog harnesses; however, this is lacking in the pet dog harness industry. This study aimed to explore the impact of six different harness types on canine biomechanics compared to a standard (base) collar, specifically analysing shoulder extension, shoulder flexion, elbow extension and elbow flexion. A high-speed recording device recorded a total of 21 videos per dog (n=30), and analysis demonstrated that the straight-front harness allowed for the most elbow and shoulder extension and flexion, whereas the front clip allowed for the least elbow and shoulder flexion and extension. By directly comparing the six harnesses, three of which are new to research inclusion, the evidence drawn demonstrates which harnesses should be utilised more and which should potentially be avoided. These results provide a foundation for future research, and recommendations have been outlined, specifically considering breed difference, morphological impacts and long-term effects of harness design on canine gait. With advancements in the canine industry, this research is crucial for maximising safety and potentially preventing or delaying various musculoskeletal disorders.
Keywords: Biomechanical impact of pet dog harnesses, Elbow and Shoulder flexion and extension, Canine gait during harness use, Canine pet harnesses, Quintic Software for canine gait analysis, Restrictive vs. non-restrictive harnesses
The use of harnesses for pet, working and sport dogs is becoming increasingly popular in the United Kingdom (Cobb et al., 2015; Grainger et al., 2016; Shih et al., 2021). Research has explored the use of harnesses on guide dogs (Peham et al., 2013), and the behavioural impacts of harnesses (Grainger et al., 2016; Kumpulainen et al., 2021); however, there is a noticeable lack of research focusing on harnesses for pet dogs. Harness brands Ruffwear (2020) and Julius K9 (2020b) have conducted research to demonstrate the benefits of their own harnesses, however there are no comparisons between brands. There is a risk that research conducted by parties with financial investment may withhold data to maintain a positive public image (Fabbri et al., 2018), which therefore lowers the reliability of their findings.
Nagymáté et al. (2018) conducted gait analysis using three Julius K9 harnesses and found no significant differences between the three harnesses when compared to unleashed movements. This directly challenges Julius K9’s reputation for causing shoulder injuries and limiting movement (Thompson, 2020). Julius K9 (2020a) states that this opinion is caused by a lack of research and misinformation spread on social media.
Pálya et al. (2022) completed a follow-up study investigating two ‘restrictive’ harnesses and two ‘non-restrictive’ harnesses, including three Julius K9-branded harnesses. They could not recommend one particular harness, as the results gathered depended on the breed and use of the harness (Edmunds et al., 2021). This further shows that harness impact depends mainly on the build of the individual dog, suggesting that opinions on certain brands should not be generalised.
Zink (2019) hypothesised that a straight-front harness would restrict movement more than a y-front harness, but found the opposite. They assessed gait at walking speed and trot speed, and found that the y-front allowed for less shoulder extension than the straight front. They assumed this to be caused by incorrect harness fitting and is therefore a limitation of the study as other research has shown that ill-fitting harnesses can impact gait (McMillan and Spaulding, 2022).
Lafuente et al. (2018) completed a similar study utilising a treadmill but implemented markers on the dogs for angle analysis. They also found y-shaped harnesses reduced shoulder extension more than the straight front. While markers can increase accuracy (Moro et al., 2022), studies have shown that markers can influence the dogs’ natural motion due to the treadmill pressure increasing skin movement (Schwencke et al., 2012). This means that markers may lead to inaccurate results due to the causation of extraneous variables.
Söhnel et al. (2022) advocate for treadmill-based gait analysis due to its speed consistence and reduced movement variability compared to overground locomotion. However, Piccione et al. (2012) expressed that minimal treadmill experience may impact a dogs’ behaviour, therefore influencing their gait. Treadmills only allow for the joints to perform certain movements (Torres et al., 2013), which could potentially alter their gait (McIntyre, 2019). This may misrepresent the impact of harnesses on typical pet usage.
O’Neill et al. (2017) identified that musculoskeletal disorders, such as osteoarthritis, are a leading cause of death in German Shepherd Dogs. Mocchi et al. (2020) found extensive reports of osteoarthritis in veterinary practices. Osteoarthritis is the irreversible, progressive degeneration of bone and cartilage (Zeira et al., 2018). Belshaw et al. (2020) looked into gait changes caused by osteoarthritis, finding that exercise was severely impacted. Considering the influence of a harness on the musculoskeletal system, it is likely that dogs will adapt their gait to accommodate a harness during exercise, and this adaptation may be more severe if the harness is ill-fitting (Anderson, 2020).
Harnesses can alter posture, causing asymmetries between specific joints (Packer and Tivers, 2015). This can significantly impact breeds with conformational disorders, such as English Bulldogs (Escobar et al., 2017). Conformational disorders occur in breeds with exaggerated body structures, causing negative impacts to their health and welfare (Packer et al., 2023). These body structures have been shown to influence the movement of different limbs and joints due to the predisposition of orthopaedic diseases and the interference with locomotion, causing changes in asymmetry and peak vertical force (Escobar et al., 2017; Humphries et al., 2020a; Humphries et al., 2020b; Jeandel and Garosi, 2018).
Injury can lead to gait abnormalities due to compensating to reduce weight-bearing on the affected limb(s) (Carapeba et al., 2016; Goldner et al., 2018). Żuk and Księżopolska-Orłowska (2015) investigated arthritis in children, finding a reduced range of motion in the affected limb, and microtrauma in the unaffected limb. Although this research investigated children, Meeson et al. (2019) recently discovered that both dogs and humans share commonalities in osteoarthritis, such as areas of development, and causes of development. Due to this, it is likely that dogs would show similar gait abnormalities caused by the compensation.
Additionally, equine research has established that lameness causes compensatory weight redistribution onto the non-affected limbs, resulting in gait abnormalities due to the transfer of vertical force (Bragança et al., 2020; Clayton, 2016; Maliye and Marshall, 2016). Given the musculoskeletal similarities between horses and dogs, similar compensatory redistribution is expected in dogs (Ahmed et al., 2019).
Many trainers and behaviourists recommend the use of collars over harnesses due to the misconception that harnesses actively cause a dog to pull (Landsberg et al., 2023). This is incorrect; harnesses were designed to enable sled dogs to pull (Ramey et al., 2022), but they do not inherently cause pulling (Shih et al., 2020).
Townsend et al. (2020) stated that collars can cause tracheal and oesophageal issues, especially in brachycephalic breeds. Harnesses can prevent this by redistributing pressure across the body (Carter et al., 2020; Hunter et al., 2019). However, Shih et al. (2021) found this enabled significantly more pulling, so headcollars may be more suitable for these dogs (Grainger et al., 2016). The AKC (2021) express that harnesses help to avoid injury and discourage pulling but have a greater escape risk. Both Takáčová et al. (2021) and Ward (2021) state that this is usually due to user fitting error, rather than harness design, so still recommends their use.
Despite the vast array of harness designs available, comparative biomechanical research is severely limited. Blake et al. (2019) emphasised the need for direct comparisons between multiple harness designs, specifically in relation to gait. Improved understanding of the biomechanical impacts of harnesses may aid in harness production, ensuring brands offer safe and secure harnesses. Furthermore, statistical information on a wider variety of harnesses is needed so owners can make informed decisions by providing results on the impact each style/design can have on biomechanics.
This study will close this gap by investigating the influence of harness design on shoulder and elbow biomechanics, specifically flexion and extension, in pet dogs. This new approach may facilitate the further study of the long-term effects of harnesses. This research will provide the vital information needed for owners to choose harnesses and to bring awareness to the variety of styles available.
Canines have various morphologies depending on their breed (Bannasch et al., 2021). Improving owner awareness of the variety of harnesses available means they are more adept to finding a well-fitting, suitable harness. This can therefore prevent the risk of harnesses having a negative impact on the dogs’ biomechanics (Preston et al., 2012).
Data collection occurred at the University Centre Reaseheath (UCR), utilising various indoor classrooms to avoid weather-related disruptions (Edwards et al., 2018). Advertisement of the study was released on social media in an attempt to achieve the desired sample size (n=30). This surpasses similar research, and provided sufficient data within the time constraints (Morse, 2015). Ethical approval was obtained on 25 June 2022, before data collection began, from the UCR Ethics Committee, in accordance with the Animal Welfare Act 2006 (ethics approval reference 221405LDsub4).
The study was open to all dogs, excluding the four banned breeds in the UK (Pitbull Terrier, Japanese Tosa, Dogo Argentino and Fila Brasileiro; The Dangerous Dogs Act, 1991). Due to accepting all breeds, one harness of each size, per harness type, was required. Dogs had to be over the age of 18 months to prevent the open growth plates of puppies causing abnormal results (Virag et al., 2022). They also had to be fit and well, with no existing musculoskeletal disorders, due to these impacting biomechanics (Adrian et al., 2019) or causing harm/discomfort (Shih et al., 2021).
Participants were recruited through social media advertisement explaining the study and inviting participation. Staff who enrolled their dogs in the university kennels were also approached regarding the inclusion of their dogs. These were included due to the convenient access to the dogs during the working day.
Potential participants were required to complete a pre-trial questionnaire per dog before their trial could begin. Three screening questions were included to ensure three main requirements. The first being that the owner was at least 18 years old (UK General Data Protection Regulation, 2021). Secondly, that the dog was over 18 months old. Finally, ensuring the dog had no known musculoskeletal disorders.
The remainder of the questionnaire gathered information pertaining to each participant, such as breed and age. All owners were asked if their dog was likely to become distressed during harness fitting. If yes, the participant was excluded to ensure researcher safety (Thompkins et al., 2016).
Video data were collected using a high-speed 60fps (frames per second) camera (iPhone 11), for simplicity and availability. Each dog was fitted with a collar as a control measure (Simmons et al., 2015), and each harness was fitted by, or under the guidance of, the author. This was to ensure they were correctly fitted and sized to avoid either factor becoming an extraneous variable and influencing results (Bremhorst et al., 2018).
Each dog was walked down a 4m (metre) walkway on a 1.3m lead, marked using a 5m lead, with a camera situated 1–2m away on a stand. This distance was altered depending on the height of the dog. Three repetitions per harness/collar were recorded; any unsuccessful videos were re-recorded. Each video was then transferred onto the study laptop via Telegram and saved in a password-protected Microsoft Vault.
While all participants were trialled in the same harnesses, the order of these was randomised. This may have negated the impacts of habituation to harness wear or trial behaviour expectations (Suresh, 2011). This is especially relevant for dogs who do not normally wear harnesses. Future research should consider the benefits of randomised studies and/or training prior to the trial.
The independent (explanatory) variable was the harness style (see Appendix 1), and the dependent (response) variable were elbow extension, elbow flexion, shoulder extension and shoulder flexion (see Appendices 2 and 3). This study included a third variable, a (random) data variable, which was the dogs involved. As a control measure, each dog was walked on the base collar to establish an approximate base-gait measurement. Owners picked the direction of walking, and this remained constant for each video to ensure directional changes had no impact.
Videos were then analysed using Quintic Biomechanics Software (Quintic Software, 2009), due to its ability to provide high-speed video capture and its allowance for biomechanical analysis. For each video, the gait cycle was broken into three phases: (1) full elbow extension/shoulder flexion (as we can measure the different angles on the same frame), (2) elbow flexion and (3) shoulder extension (see Appendices 2 and 3). These were then measured using Quintic Manual Angular Software.
Each angle was then exported onto a large Excel document for each walk. The mean and R Studio were used to fully analyse the data. Primary data was collected; although this can be more expensive and time-consuming to complete, data gathered first-hand has increased reliability and validity (Vetter, 2017).
A Generalised Linear Mixed Model (GLMM) was used for data analysis, due to being appropriate for large amounts of clustered data (Rabe-Hesketh and Skrondal, 2010); this study collected approximately 2500 individual pieces of data. As previously noted, three different walks were recorded per dog per harness per angle. From there, each individual piece of data was put onto a final data set, before being uploaded onto R Studio for GLMM analysis. Outliers have not been removed, as GLMMs are flexible and can handle random effect and non-normal distributions, so this step was not necessary (Bolker et al., 2009). Alongside this, the median and interquartile range were found for each harness type and angle (as shown in Table 1).
Due to the third, random variable – the dogs themselves – the GLMM accounted for this by basing itself on the entire sample population instead of each individual subject, which allowed for sparse sampling (Huang and Li, 2007). This is why the raw data set was used, as GLMM works best with large sets of data instead of means, and it can appropriately incorporate/exclude any outliers.
A normality test is not required for a GLMM due to the large amount of data it handles and the assumption that there will be non-normal results within the data collected (Stroup, 2015). However, an omnibus normality test (Shapiro-Wilk) was still performed to assess the normality of the whole data set (p=0.05) and to justify the correct descriptive statistics to report. The data was not normally distributed (p<0.05), so the median and interquartile range were used to remain consistent across the data.
Alongside this, 28 individual Shapiro-Wilkes normality tests were performed for each harness/collar and angle combination (elbow extension/straight front, elbow extension/chest plate, for example – see Appendix 4) as supplementary information. Although some of these normality tests were normal, if there is any non-normal data then the assumption that the entire dataset is non-normal is made (Field, 2017).
However, these normality tests were used to decide which pairwise tests to use for the result comparisons. A one-way ANOVA was used for shoulder flexion as all results were normal, and a Kruskal Wallis was used for the other three angles, as the results were a mixture of normal and non-normal.
The health and safety, and ethics for this study were upheld, meaning this research was ethically produced, and demonstrates to future researchers how to ethically conduct dog-related studies (Woodin, 2015). Particularly, one dog was excluded from the trial due to their negative reaction to the study location, so removing him from the study was the most ethical decision (Tasker et al., 2018).
Alongside this, participants never came into contact with other dogs outside of their household (King and Zohny, 2022). Timeslots were given to each participant, allowing sufficient time for them to leave before the next participant’s trial. This ensured that dogs did not come into contact with each other; to give us time to finish each harness; and to ensure the dog had time to acclimatise to the room before beginning the trial.
Any dogs with nervousness or disfavour of having their paws touched were handled by the owner to protect researchers from any reactions (Health and Safety at Work Act 1974), but to also keep anxiety levels of the dogs as low as possible (Animal Welfare Act 2006). Although changing the walker for some dogs may be an extraneous variable, the author prioritised the ethics and safety of each participant as paramount (National Dog Warden Association, 2012).
A GLMM was performed to compare the influence of each harness on elbow and shoulder extension and flexion. Descriptive statistics are presented in Table 1, with results expressed as medium and interquartile range.
Results are recorded as (median (IQR: 25th percentile – 75th percentile)).
The influence of each harness design on shoulder and elbow extension and flexion, compared to the base collar is displayed. The descriptive statistics are recorded as (median (IQR 25th percentile – 75th percentile)).
As indicated below in Figure 1, the front clip (132.1° (IQR: 127.1° – 135.2°)) and no-pull (133.2° (IQR: 128.3° – 136.7°)) had the most significant impact on elbow extension compared to the base collar (137.0° (IQR: 131.8° – 141.4°)). The chest plate (134.2° (IQR: 130.5° – 138.3°)) and step-in (133.9° (IQR: 129.4° – 139.3°)) both had a significant impact, but less so than the aforementioned harnesses. However, neither the straight front (134.6° (IQR: 130.1° – 140.4°)) or y-front (135.9° (IQR: 130.8° – 139.7°)) had a significant impact on elbow extension.
The degree of canine elbow extension for each harness design and base collar is displayed (n=60). This is displayed as a box-and-whisker plot with outliers.
(GLMM: chest plate t = -0.942, p=0.034; front clip t = -2.123, p<0.001; no-pull t = -5.642, p<0.001; step-in t = -4.057, p=0.027; straight front t = -2.210, p=0.078; y-front t = -1.765, p=0.346).
Conveyed in Figure 2, the front clip (76.4° (IQR: 69.6° – 81.0°)) and step-in (75.7° (IQR: 71.0° – 80.5°)) had the most significant impact on elbow flexion compared to the base collar (80.9° (IQR: 73.7° – 86.9°)). The no-pull (78.0° (IQR: 74.0° – 83.1°)) also had a significant impact, but not to a considerable degree. The straight front (79.7° (IQR: 75.9° – 84.3°)), y-front (79.5° (IQR: 75.2° – 85.0°)) and chest plate (81.4° (IQR: 74.0° – 84.2 °)) did not have a significant impact on elbow flexion. (See Appendix 5 for the pairwise comparisons).
The degree of canine elbow flexion for each harness design and base collar is displayed (n=60). This is displayed as a box-and-whisker plot with outliers.
(GLMM: chest plate t = -0.755, p=0.511; front clip t = -0.658, p<0.001; no-pull t = -5.769, p=0.003; step-in t = -3.006, p<0.001; straight front t = -5.471, p=0.501; y-front t = -0.673, p=0.451).
Exhibited in Figure 3, the chest plate (88.5° (IQR: 80.9° – 97.6°)), front clip (84.7° (IQR: 79.6° – 95.2°)), no-pull (88.2° (IQR: 76.8° – 95.2°)), step-in (86.3° (IQR: 80.9° – 96.2°)), straight front (93.8° (IQR: 86.0° – 99.6°)) and y-front (93.5° (IQR: 84.3° - 99.3°)) all had a significant impact on shoulder extension compared to the base collar (97.4° (IQR: 88.5° – 102.5°)). (See Appendix 6 for the pairwise comparisons).
The degree of canine shoulder extension for each harness design and base collar is displayed (n=60). This is displayed as a box-and-whisker plot with outliers.
(GLMM: chest plate t = -3.863, p<0.001; front clip t = -7.403, p<0.001; no-pull t = -9.514, p<0.001; step-in t = -10.030, p<0.001; straight front t = -8.340, p<0.001; y-front t = -3.547, p<0.001).
Displayed in Figure 4, the chest plate (88.4° (IQR: 82.5° – 94.5°)), front clip (84.1° (IQR: 78.2° – 90.7°)), no-pull (85.7° (IQR: 79.7° – 91.2°)), step-in (87.6° (IQR: 82.8° – 93.8°)) and y-front (88.9° (IQR: 84.4° – 93.8°)) all had a significant impact compared to the base collar (91.3° (IQR: 85.9° – 97.6°)). The only harness to not significantly impact shoulder flexion was the straight front (89.6° (IQR: 85.4° – 94.9°)). (See Appendix 7 for the pairwise comparisons).
The degree of canine shoulder flexion for each harness design and base collar is displayed (n=60). This is displayed as a box-and-whisker plot with outliers.
(GLMM: chest plate t = -3.557, p<0.001; front clip t = -4.736, p<0.001; no-pull t = -9.184, p<0.001; step-in t = -7.824, p<0.001; straight front t = -5.079, p=0.02; y-front t = -2.330, p<0.001).
Across all four analyses, the front clip is the only harness that significantly reduced all four angles measured. Opposingly, the straight-front harness did not significantly reduce three of the angles measured. The only significant reduction was of shoulder extension; however, every harness significantly reduced this angle compared to the collar. The y-front did not significantly reduce elbow extension or flexion but did significantly reduce shoulder extension and flexion. (See Appendix 8 for the ANOVA and Tukey honest significant difference post-hoc).
The chest plate did not reduce elbow flexion, but significantly reduced both shoulder extension and flexion. It also reduced elbow extension, but to a lower statistical degree. Finally, the no-pull and step-in both reduced every angle, with both significantly reducing shoulder extension and flexion. Both harnesses reduced elbow flexion, but the reduction was more significant in the step-in. Similarly, both also reduced elbow extension, but the reduction was more significant in the no-pull.
Alongside reporting the median and interquartile range, the mean and standard deviation (mean +/- standard deviation) were noted, as presented in Table 2 below. This table displays the influence of each harness design on shoulder and elbow extension and flexion. The coloured shadings are explained in Figure 5. Furthermore, a post-hoc test was performed for each angle to show more clearly where the differences lie between each harness/collar, and which are significant.
This key displays the harnesses in order of most extension/flexion measured. The base collar is not included.
In Table 2, the base collar was not included in the highlighted order of flexion/extension allowed, as this was expected to be higher due to the absence of contact with the shoulder/elbow. Table 2 also includes the maximum flexion and extension of the shoulder and elbow from Lorke et al.’s (2017) study on Beagle dogs for comparison and accuracy assurance.
In this study, a comprehensive analysis was performed using several statistical methods, including a GLMM, three Kruskal Wallis with pairwise comparisons and a one-way ANOVA with post-hoc tests. Comparing the results, they differ, which is most likely due to the strengths of these tests and their abilities to work with outliers. Each test was performed to provide valuable insights; however, due to the flexibility and accuracy of the GLMM when handling large amounts of complex data, the discussion will focus on the results from the GLMM.
Reviewing the results of the GLMM, all six harnesses significantly reduced shoulder extension compared to the base collar. This finding disproves the hypothesis that there would be a significant difference in shoulder extension between the y-front and front-clip harness. However, because every harness impacted this angle, future research could remove the base collar to determine which harness has the least impact, as a collar may not always be an option for an owner (Bailey, 2022; Bolton et al., 2021).
Exploring the front-clip harness, previously considered ‘non-restrictive’, (Zink, 2019) this harness significantly reduced all four angles measured (EE (132.1°), EF (76.4°), SE (84.7°), SF (84.1°)). The y-front harness also significantly reduced shoulder extension (SE) (93.5°) and flexion (SF) (88.9°) but had minimal impact on elbow extension (EE) (135.9°) and flexion (EF) (79.5°). Blake et al., (2019) deem this impact to be caused by the y-front restricting the scapula angulation (Aspinall and Cappello, 2019), thus reducing movement.
Both the front-clip and y-front harnesses are manufactured with the strap running from the sternum to the dorsal neck, which could potentially influence biomechanics due to the loading on the sternum (Peham et al., 2013). More research would be required to compare strap positioning and sternum impact. This harness design is thought of as being ‘non-restrictive’ due to the strap not crossing the chest; however, future research would be necessary to definitively prove which strap positioning has the least impact on biomechanics.
When comparing the elbow extension and flexion of the y-front and front clip, it is interesting that only one resulted in significant impact, as this potentially shows that the strap positioning is not the key factor in what influences flexion and extension. Due to the similar style, the other variables of lead position and strap width must be considered. More research is needed to focus on chest-strap width and its interference with elbow movement, as this may be the factor causing the difference in impact. The largest front-clip harness width is 4 inches at the widest part, compared to the largest y-front harness width being 1.5 inches at the widest part. Due to this large difference, it is possible that this factor would have an impact, but more research is needed to prove/disprove this.
The straight front allowed for the largest shoulder extension (93.8°) and flexion (89.6°), and the second-most elbow extension (134.6°), only significantly impacting shoulder extension (79.7°). Many studies have hypothesised that the ‘non-restrictive’ y-front harness would have less of an impact than the ‘restrictive’ straight-front harness (Pálya et al., 2022; Sandberg, 2022), yet results have shown the opposite. Zink’s (2019) study also found that the non-restrictive y-front significantly restricted shoulder extension when walking. Conclusions can be drawn that the straight front had the least impact on canine biomechanics, which is a key finding for the pet dog community as this may lessen the negative associations with this style (Julius K9, 2020b).
Furthering the comparison between straight-fronts and y-fronts, Pálya et al.’s. (2022) study compared the Julius K9 straight front to the new Julius K9 y-front. They found that both harnesses limited gait compared to a standard collar, but there was no significant difference between the two harness designs, thus showing a straight front is not necessarily ‘restrictive’ compared to a y-front. They concluded that movement restriction depends on the individual dog, and the fit of the harness, however expressed that future research would be needed to support this.
Clayton et al. (2017) completed equine research into saddle fitting, and discussed how saddles impact stride, velocity and acceleration. They stated that a poor-fitted saddle impairs the mobility of the horse’s back, and inhibits a consistent gait, which can cause lameness and musculoskeletal issues in the long term. Linking this to pet dogs, if an owner struggles to fit a harness, it can impact their gait, which can cause issues over time. Future studies could see the ability of owners to fit different harnesses, and this could encourage brands to increase the harness-fit education they provide.
This research may aid in the safe production and scientifically informed harness advertisement and information to the public. It may also help dispel the incorrect beliefs around harness restriction. As discussed, the y-front harness is commonly sold as ‘non-restrictive’, but this study has proven that it restricts flexion and extension of both the elbow and shoulder, which is a misrepresentation to the buyers. Alternatively, the Julius K9, which is often labelled as ‘restrictive’ and is regularly disfavoured by animal professionals and the public, has been proven to have little impact on canine biomechanics. This also supports Julius K9’s (2020a) statement regarding the safety and impact of their harness.
Results of this study found that the chest plate and step-in also significantly reduced shoulder extension (chest plate (88.5°), step-in (86.3°)) and flexion (chest plate (88.4°), step-in (87.6°)). However, both also significantly reduced elbow extension (chest plate (134.2°), step-in (133.9°)), but to a lesser degree. The step-in harness did significantly reduce elbow flexion (75.7°), whereas the chest plate did not (81.4°).
Focusing on the step-in harness, this harness has a similar design to the straight front, with the chest strap crossing the chest and shoulders. The Julius K9 straight front has a thicker chest strap that could have impacted gait less, due to the larger distribution of pressure, whereas the step-in chest strap is thinner, so pressure is distributed to a much smaller area. This could explain why the straight front had a much less impact on all measurements compared to the step-in. This refutes the hypothesis that predicted the step-in would have less impact than the straight front, and clearly demonstrates the need for future research to consider chest-strap width and its impact on biomechanics.
Peham et al. (2013) measured the pressure distribution of three straight-front guide dog harnesses with different chest-strap widths. They found that when the forelimb extended, the pressure force on the chest increased, but remained low when the forelimb was in flexion. They also found that the pressure force was lowest in the harness with the widest chest strap. Future research could involve similar research but using pet dogs to examine what width chest strap is optimum for low pressure force but still allows for standard flexion and extension.
Moving onto the no-pull harness; this is highly recommended for dogs that pull due to the tightening mechanism under the axilla (armpit), causing slight discomfort to the dog (Company of Animals, 2022). This harness has never been included in research, based on the authors’ knowledge. The no-pull reduced all four angles measured: it significantly reduced shoulder extension (88.2°) and flexion (85.7°), and elbow extension (133.2°) and flexion (78.0°), but it reduced elbow extension to a lesser significant degree.
Coincidentally, both the y-front and the no-pull are produced by the brand Halti (VioVet, 2023), which is interesting that their no-pull had such a significant impact on biomechanics, whereas the y-front had very little impact. This demonstrates the requirement for future research to explore the different Halti-branded harnesses and compare how each influences canine gait, potentially supporting their future harness production.
Exploring all six harnesses, the hypothesis that a significant difference in biomechanics would be witnessed depending on the harness worn has been confirmed. However, based on the statistical data, the straight front had the least impact on canine gait, therefore is the authors’ top recommendation.
As highlighted previously, there is a need for future research to focus on canine gait and harness impact (Blake et al., 2019). Lafuente et al. (2018) discussed the necessity for future research to include multiple harness styles and brands. They also discussed the requirement of studies to include a researcher with experience surrounding dogs and physiology to improve the reliability of the results (Wang et al., 2022). This study has confidently met both future research recommendations by including six harness designs and being controlled by an author with canine experience. Moreover, this study has overcome the limitations of previous studies by ensuring a larger sample size (Zink, 2019), a variety of breeds (Winter, 2013), and controlled extraneous variables, such as the influence of weather or treadmills (Lafuente et al., 2018; Söhnel et al., 2022).
The results of this study have been consistent with the results found by Lorke et al. (2017). Their data was included in Table 2 to demonstrate the maximum elbow and shoulder extension, and flexion achieved by healthy Beagle dogs during free movement. As results remained under the threshold identified by Lorke et al. (2017), and there is consistency within the ranges, this demonstrates a lack of anomalous data and suggests a high accuracy of the manual angles drawn in this study. Moreover, this suggests high concurrent validity due to the agreement between this study and Lorke et al. (2017).
Alongside this, internal consistency reliability has been assured, as results from each harness style have remained consistent (Salonen et al., 2021). Both construct and content validity have also been upheld by ensuring the software used for analysis was specifically for biomechanics (Topál et al., 2019), and that all angles were measured in the same way (Rocznik et al., 2014).
The results and conclusions drawn from this study will provide invaluable information not only for pet dog owners, but for future researchers who wish to build on these findings. However, the author acknowledges the presence of some limitations within this study in order for others to build on these and provide essential future research.
Due to the time restraints within this study, only 30 dogs were included. While Ker and Ramalingam (2013) deemed this sufficient as the sample is significantly larger than other studies (Laverack et al., 2021; Wiener and Haskell, 2016), the author identifies that a larger sample size would allow for investigation into the morphological differences between breeds (Carlisle et al., 2019; Hecht et al., 2019). This would allow for more specific harness recommendations for owners based on individual breed conformation (Voss et al., 2011).
Due to the nature of a university setting, flexibility in study site was required. The author does not feel this affected the results; however, future studies should ensure site consistency (Desai, 2020). Although this study considered harness fitting and possible reactions to the researcher in the pre-trial questionnaire, it did not preassess the dogs’ comfort level or experience of being walked on the lead by a stranger. To ensure ethicality, some dogs were walked by their owners to alleviate anxiety; however, future studies should ensure participants are comfortable with being handled by a stranger as part of the screening questions.
To help ensure reliability of results, the same researcher completed every manual angle measurement. However, as angular data was produced manually, there may be a level of unconscious or observer bias. Tanneberger and Ciupitu-Plath (2017) conclude that existing beliefs can impact the results drawn, as has been discussed above with Zink’s (2019) study. To mitigate observer bias, future studies should involve multiple observers, as well as following blinding approaches if possible, meaning the researcher measuring the angle does not know which harness they are assessing. Alternatively, other researchers who are not well-informed in the field of harness study could be involved for angle measurements, so they do not have preconceived opinions on each harness.
Alternatively, markers could be used so Quintic Automatic Angle Measurement can be used, which may increase reliability (Engelsman et al., 2022). If markers are used in future research, limitations with this technique such as lacking visibility on thick-furred dogs or movement alterations caused by the markers must be accounted for (Torres et al., 2013).
The inclusion of other gair parameters, such as stride length, would strengthen these findings further (Torres et al., 2017). Size difference and breed conformation impacts stride length, which could influence extension and flexion, so inclusion would add further clarity (Bliss et al., 2022). Furthermore, comparisons of different gaits (e.g. hindlimb angles, or dogs at a trot) would provide further results. This would be especially relevant when researching the long-term effects of harnesses, as pet dogs vary their gait during standard walks (Kano et al., 2016).
Longitudinal studies exploring the correlation between harness design and use and musculoskeletal disorders will help inform harness choices. Ethical considerations must be made here, as this study has shown the potential biomechanical influences of some harness designs (Murray et al., 2021).
A final recommendation for future researchers is the consideration of a larger array of harnesses and brands available to pet owners. Analysis of the different styles within brands (e.g. the Julius K9 straight front and y-front), different brands within styles (e.g. Ruffwear y-front and Julius K9 y-front) and differing the attachment point (e.g. front clip and back clip) would deepen the understanding of harnesses and biomechanics. Focus on brands more/less readily available may result in shifts in the market, improving the knowledge of owners and widening their harness choices.
Despite the aforementioned limitations and requirements for future research, the results of this study are still valid and beneficial for the canine industry as a whole and provides the building blocks for students and researchers to study further into this field. Providing an accessible version for owners will impart clarity and enable them to make informed decisions. Hopefully, these results will drive the future of harness production to ensure for maximum safety and comfort for all dogs.
This study has concisely demonstrated that the choice of pet dog harness impacts biomechanics, closing the research gap into harness impact on canine flexion and extension in pet dogs. The results display the impact of each via angulation and have demonstrated how different harness styles impact shoulder and elbow extension and flexion.
Comparing results of all six harnesses, this study can confidently recommend the Julius K9 straight-front harness or the Halti y-front harness as the preferred option for pet owners. Furthermore, the highly regarded Ruffwear front-clip harness exhibited the largest impact on joint flexion and extension overall, thus illustrating its unsuitability. The author notes that while these harnesses were graded on suitability in results, the aim was not to conclude ‘the best’ harness overall; owners must use the information to select the best-suited harness for their individual dog.
Careful analysis of previous studies has allowed this study to accurately improve on previous limitations and draw updated conclusions on their findings. The study has demonstrated a need for further research to assess the impact of harness width on canine biomechanics. Research into breed differences will also help expand this area further.
These findings provide long-term benefits to the pet dog community. The inclusion of commonly available harnesses means that these conclusions are accessible for the average dog owner, and will help them to make informed decisions on their dogs’ safety and wellbeing. This also provides appropriate alternatives and reassurance for those dogs for whom a collar is unsuitable.
I would like to acknowledge and thank the following individuals who supported me throughout my research:
My supervisor, Lucianna Churchill, for supporting me (and putting up with my endless questions!) throughout this project. To Paige McCormack for helping with data collection and sourcing participants.
My partner, Ryszard Bialosiewicz, for the moral support and endless cups of coffee provided. To my assistance dog Tucker, for keeping me company and making university more manageable.
To my good friend Leah Sullinge, for always telling me to follow my dreams and encouraging me to begin my career in the canine world.
Companies’ Ruffwear and EzyDog, for supplying harnesses, and a special thanks to Mandy Taylor and Ashleigh King from Pets at Home for your guidance on harnesses, and for supplying a large number of them.
My classmates Natalia Szczerba, Rosanna Frith, Isabel Oborne and Jake Clements for helping me during the data collection.
A big thank you to Abigail Hughes, for the hours spent on FaceTime, proof-reading, and always offering a helping hand!
Finally, to the wonderful owners of the dogs included in my study!
Figure 1: Box-and-whisker plot showing elbow extension for each harness design
Figure 2: Box-and-whisker plot showing elbow flexion for each harness design
Figure 3: Box-and-whisker plot showing shoulder extension for each harness design
Figure 4: Box-and-whisker plot showing shoulder flexion for each harness design
Figure 5: The highlighting key for Table 2
Table 1: Median and interquartile range of each harness design and base collar
Table 2: Mean and standard deviation of each harness design and base collar
Table A1: Harness brands, names, styles and images.
Table A2: The four gaits measured
Table A3: The four gaits measured
Table A4: Normality tests performed for each harness/angle combination
Table A5: Comparisons between each collar/harness type using a Kruskal Wallis with pairwise (chi-squared = 28.57, df = 6, p<0.001)
Table A6: Comparisons between each collar/harness type using a Kruskal Wallis with pairwise (chi-squared = 42.87, df = 6, p<0.001)
Table A7: Comparisons between each collar/harness type using a Kruskal Wallis with pairwise (chi-squared = 50.89, df = 6, p<0.001)
Table A8: Comparisons between each collar/harness type using a Tukey’s HSD test (F = 9.02, df = 6, p<0.001)
Table A1: Harness brands, names, styles and images.
| Harness Brand | Harness Name | Harness Style | Image of Harness |
| Julius K9 | IDC© Powerharness | Straight front |
|
| Ruffwear | Front Range© Dog Harness | Front-range |
|
| EzyDog | Chest Plate Harness | Chest plate |
|
| Halti | Comfort Collar | Base collar |
|
| No Pull Harness | No-pull |
|
|
| Walking Harness | Y-front |
|
|
| 3 Peaks | Step-In Harness | Step-in |
|
Table A2: The four gaits measured
Table A3: The four gaits measured
This supplementary material provides the results from all 28 normality tests performed for each harness/angle combination. Although an omnibus normality test was performed overall, the individual results are concluded for additional information. The following table summarises the test statistics and p-values for each combination (all were analysed using a Shapiro-Wilk normality test).
Table A4: Normality tests performed for each harness/angle combination
| Angle | Harness | Test Statistic (W) | P-Value | Comments |
| Elbow Extension | Base Collar | 0.96129 | p<0.01 | Non-normal |
| Elbow Extension | Straight Front | 0.96393 | p<0.05 | Non-normal |
| Elbow Extension | Front Clip | 0.98338 | p>0.1 | Normal |
| Elbow Extension | Chest Plate | 0.98373 | p>0.1 | Normal |
| Elbow Extension | Y-Front | 0.97280 | p>0.05 | Normal |
| Elbow Extension | No-Pull | 0.98833 | p>0.1 | Normal |
| Elbow Extension | Step-In | 0.98822 | p>0.1 | Normal |
| Elbow Flexion | Base Collar | 0.98283 | p>0.1 | Normal |
| Elbow Flexion | Straight Front | 0.98947 | p>0.1 | Normal |
| Elbow Flexion | Front Clip | 0.98160 | p>0.1 | Normal |
| Elbow Flexion | Chest Plate | 0.96669 | p<0.05 | Non-normal |
| Elbow Flexion | Y-Front | 0.98364 | p>0.1 | Normal |
| Elbow Flexion | No-Pull | 0.97726 | p>0.1 | Normal |
| Elbow Flexion | Step-In | 0.97162 | p<0.05 | Non-normal |
| Shoulder Extension | Base Collar | 0.98547 | p>0.1 | Normal |
| Shoulder Extension | Straight Front | 0.98047 | p>0.1 | Normal |
| Shoulder Extension | Front Clip | 0.97633 | p>0.05 | Normal |
| Shoulder Extension | Chest Plate | 0.97865 | p>0.1 | Normal |
| Shoulder Extension | Y-Front | 0.96099 | p<0.01 | Non-normal |
| Shoulder Extension | No-Pull | 0.97347 | p>0.05 | Normal |
| Shoulder Extension | Step-In | 0.98783 | p>0.1 | Normal |
| Shoulder Flexion | Base Collar | 0.98368 | p>0.1 | Normal |
| Shoulder Flexion | Straight Front | 0.98531 | p>0.1 | Normal |
| Shoulder Flexion | Front Clip | 0.98815 | p>0.1 | Normal |
| Shoulder Flexion | Chest Plate | 0.97574 | p>0.05 | Normal |
| Shoulder Flexion | Y-Front | 0.98522 | p>0.1 | Normal |
| Shoulder Flexion | No-Pull | 0.99054 | p>0.1 | Normal |
| Shoulder Flexion | Step-In | 0.98246 | p>0.1 | Normal |
The normality tests indicate that although the majority of the data had a normal distribution, some were non-normally distributed showing significant deviations.
Due to there being non-normal data in the elbow extension results, a non-normal pairwise test is required, which is why a Kruskal Wallis with pairwise has been chosen. This shows comparisons between each collar/harness type.
Table A5: Comparisons between each collar/harness type using a Kruskal Wallis with pairwise (chi-squared = 28.57, df = 6, p<0.001)
| Base Collar | Str. Front | Front Clip | Chest Plate | Y-Front | No-Pull | |
| Str. Front | 1.00000 | - | - | - | - | - |
| Front Clip | 0.00295 | 0.00507 | - | - | - | - |
| Chest Plate | 1.00000 | 1.00000 | 0.00141 | - | - | - |
| Y-Front | 1.00000 | 1.00000 | 0.00339 | 1.00000 | - | - |
| No-Pull | 0.48788 | 0.74240 | 0.48788 | 0.48788 | 0.73307 | - |
| Step-In | 0.00166 | 0.00082 | 1.00000 | 0.00022 | 0.00070 | 0.17544 |
Due to there being non-normal data in the elbow flexion results, a non-normal pairwise test is required, which is why a Kruskal Wallis with pairwise has been chosen. This shows comparisons between each collar/harness type.
Table A6: Comparisons between each collar/harness type using a Kruskal Wallis with pairwise (chi-squared = 42.87, df = 6, p<0.001)
| Base Collar | Str. Front | Front Clip | Chest Plate | Y-Front | No-Pull | |
| Str. Front | 1.00000 | - | - | - | - | - |
| Front Clip | 0.00073 | 0.05569 | - | - | - | - |
| Chest Plate | 0.54035 | 1.00000 | 0.07444 | - | - | - |
| Y-Front | 1.00000 | 1.00000 | 0.00307 | 1.00000 | - | - |
| No-Pull | 0.01488 | 0.54035 | 1.00000 | 0.87149 | 1.00000 | - |
| Step-In | 0.63946 | 1.00000 | 0.23166 | 1.00000 | 1.00000 | 1.00000 |
Due to there being non-normal data in the shoulder extension results, a non-normal pairwise test is required, which is why a Kruskal Wallis with pairwise has been chosen. This shows comparisons between each collar/harness type.
Table A7: Comparisons between each collar/harness type using a Kruskal Wallis with pairwise (chi-squared = 50.89, df = 6, p<0.001)
| Base Collar | Str. Front | Front Clip | Chest Plate | Y-Front | No-Pull | |
| Str. Front | 0.20795 | - | - | - | - | - |
| Front Clip | 5.4e-06 | 0.00570 | - | - | - | - |
| Chest Plate | 0.00078 | 0.22826 | 1.00000 | - | - | - |
| Y-Front | 0.20795 | 1.00000 | 0.04041 | 0.51661 | - | - |
| No-Pull | 4.3e-06 | 0.01594 | 1.00000 | 1.00000 | 0.02093 | - |
| Step-In | 1.2e-05 | 0.02711 | 1.00000 | 1.00000 | 0.12960 | 1.00000 |
Shoulder flexion had normal data for all harnesses/collars. Due to this, a normal pairwise test is required, which is why a One-Way ANOVA with Tukey’s HSD test has been chosen. This shows comparisons between each collar/harness type.
Table A8: Comparisons between each collar/harness type using a Tukey’s HSD test (F = 9.02, df = 6, p<0.001)
| Base Collar | Str. Front | Front Clip | Chest Plate | Y-Front | No-Pull | |
| Str. Front | 0.60419 | - | - | - | - | - |
| Front Clip | 0.00000 | 0.00016 | - | - | - | - |
| Chest Plate | 0.01801 | 0.69844 | 0.05449 | - | - | - |
| Y-Front | 0.15977 | 0.98634 | 0.00411 | 0.98594 | - | - |
| No-Pull | 0.00000 | 0.00437 | 0.98472 | 0.33900 | 0.05646 | - |
| Step-In | 0.00898 | 0.56080 | 0.09482 | 0.99999 | 0.95470 | 0.46886 |
Adrian, C., K. Haussler, C. Kawcak, R. Reiser, C. Riegger-Krugh, R. Palmer, C. McIlwraith and R. Taylor (2019), ‘Gait and Electromyographic alterations due to early onset of injury and eventual rupture of the cranial cruciate ligament in dogs: A pilot study’, Veterinary and Comparative Orthopaedics and Traumatology, 48 (3), 388–00
Ahmed, W., M. Kulikowska, T. Ahlmann, L. Berg, A. Harrison and V. Elbrønd (2019), ‘A comparative multi-site and whole-body assessment of fascia in the horse and dog: A detailed histological investigation’, Journal of Anatomy, 235 (6), 1065–77
AKC. (2021), ‘Is a dog harness better than a dog collar?’, available at https://www.akc.org/expert-advice/training/dog-harness-vs-collar/, accessed 12 April 2023
Anderson, K., H. Zulch, D. O’Neill, R. Meeson and L. Collins (2020), ‘Risk factors for canine osteoarthritis and its predisposing arthropathies: A systematic review’, Frontiers in Veterinary Science, 7, 220
Aspinall, V. and M. Cappello (2019), Introduction to Animal and Veterinary Anatomy and Physiology (4th ed.), Wallingford: England: CABI
Bailey, J. (2022), ‘An investigation of force potential against the canine neck associated with collar use’, ProQuest, 29061866
Bannasch, D., T. Famula, J. Donner, H. Anderson, L. Honkanen, K. Batcher, N. Safra, S. Thomasy and R. Rebhun (2021), ‘The effect of inbreeding, body size and morphology on health in dog breeds’, Canine Medicine and Genetics, 8 (1), 12
Belshaw, Z., R. Dean and L. Asher (2020), ‘Slower, shorter, sadder: A qualitative study exploring how dog walks change when the canine participant develops osteoarthritis’, BMC Veterinary Research, 16 (1), 85
Blake, S., R. Williams and R. Ferro de Godoy (2019), ‘A systematic review of the biomechanical effects of harness and head-collar use in dogs’, bioRxiv, https://doi.org/10.1101/759258
Bliss, M., J. Terry and R. Ferro de Godoy (2022), ‘Limbs kinematics of dogs exercising at different water levels on the underwater treadmill’ Veterinary Medical Science, 8 (6), 2374–81
Bolker, B., M. Brooks, C. Clark, S. Geange, J. Poulsen, H. Stevens and J. White (2009), ‘Generalized linear mixed models: A practical guide for ecology and evolution’, Trends in Ecology & Evolution, 24 (3), 127–35
Bolton, S., N. Cave, N. Cogger and G. Colborne (2021), ‘Use of a collar-mounted triaxial accelerometer to predict speed and gait in dogs’, Animals, 11 (5), 1262
Bragança, F., E. Hernlund., M. Thomsen., N. Waldern., M. Rhodin., A. Byström., P. Weeren and M. Weishaupt, M. A. (2020), ‘Adaptation strategies of horses with induced forelimb lameness walking on a treadmill’, Equine Veterinary Journal, 53(3), 600–11
Bremhorst, A., P. Mongillo, T. Howell and L. Marinelli (2018), ‘Spotlight on assistance dogs – Legislation, welfare and research’, Animals, 8 (8), 129
Carapeba, G., P. Cavaleti, G. Nicácio, R. Brinholi, R. Giuffrida and R. Cassu (2016), ‘Intra-articular hyaluronic acid compared to traditional conservative treatment in dogs with osteoarthritis associated with hip dysplasia’, Evidence-Based Complementary and Alternative Medicine, 2076921, 1–10
Carlisle, L., E. Memili, R. Linford, K. Slater and M. Nicodemus (2019), ‘Relationship between gait mechanics and the speed of the trot in the Weimaraner dog breed’, Topics in Companion Animal Medicine, 35, 26–30
Carter, A., D. McNally and A. Roshier (2020), ‘Canine collars: An investigation of collar type and the forces applied to a simulated neck model’, Veterinary Record, 187 (7), e52
Clayton, H. (2016), ‘Horse species symposium: Biomechanics of the exercising horse’, Journal of Animal Science, 94 (10), 4076–86
Clayton, H and S. Hobbs (2017), ‘The role of biomechanical analysis of horse and rider in equitation science’, Applied Animal Behaviour Science, 190, 123–32
Cobb, M., N. Branson, P. McGreevy, A. Lill and P. Bennett (2015), ‘The advent of canine performance science: Offering a sustainable future for working dogs’, Behavioural Processes, 110, 96–104
Company of Animals. (2022), ‘Halti no pull harness’, available at https://companyofanimals.com/uk/brand-product/halti-no-pull-dog-harness/, accessed 23 March 2023
Desai, M. (2020), ‘Recruitment and retention of participants in clinical studies: Critical issues and challenges’, Perspectives in Clinical Research, 11 (2), 51–53
Edmunds, G., M. Smalley, S. Beck, R. Errington, S. Gould, H. Winter, D. Brodbelt and D. O’Neill (2021), ‘Dog breeds and body conformations with predisposition to osteosarcoma in the UK: A case-control study’, Canine Medicine and Genetics, 8 (1) https://doi.org/10.1186/s40575-021-00100-7
Edwards, G., M. Azain and A. Parks (2018), ‘Agricultural animals as biomedical models: Occupational health and safety considerations’, ILAR Journal, 59 (2), 161–67
Engelsman, D., T. Sherif, S. Meller, F. Twele, I. Klein, A. Zamansky and H. Volk (2022), ‘Measurement of canine ataxic gait patterns using body-worn smartphone sensor data’, Frontiers in Veterinary Science, 9, 912253
Escobar, A., A. Souza., A. Fonseca and J. Matera (2017), ‘Kinetic gait analysis in English Bulldogs’, Acta Veterinaria Scandinavica, 59, 77
Fabbri, A., A. Lai, Q. Grundy and L. Bero (2018), ‘The influence of industry sponsorship on the research agenda: A scoping review’, American Journal of Public Health, 108 (11), 9–16
Field, A. (2017). Discovering statistics using IBM SPSS statistics. (5th edn.), Los Angeles, United States of America: Sage Publications ltd.
Goldner, B., S. Fischer, I. Nolte and N. Schilling (2018), ‘Kinematic adaptions to induced short-term pelvic limb lameness in trotting dogs’, BMC Veterinary Research, 14 (1), 183
Grainger, J., A. Wills and V. Montrose (2016) ‘The behavioural effects of walking on a collar and harness in domestic dogs (Canis familiaris)’, Journal of Veterinary Behaviour, 14, 60–64
Hecht, E., J. Smaers, W. Dunn, M. Kent, T. Preuss and D. Gutman (2019), ‘Significant neuroanatomical variation among domestic dog breeds’, The Journal of Neuroscience, 39 (39), 7748–58
Huang, X and J. Li (2007), ‘Pharmacokinetic-pharmacodynamic modeling and simulation’, American Journal of Pharmaceutical Education, 71 (1), 18
Humphries, A., A. Shaheen and C. Gómez Álvarez (2020a), ‘Biomechanical comparison of standing posture and during trot between German shepherd and Labrador retriever dogs’, PLOS ONE, 15 (10), e0239832
Humphries, A., A. Shaheen and C. Gómez Álvarez (2020b), ‘Different conformations of the German shepherd dog breed affect its posture and movement’, Scientific Reports, 10 (1), 1–16
Hunter, A., S. Blake and R. De Godoy (2019), ‘Pressure and force on the canine neck when exercised using a collar and leash’, Veterinary and Animal Science, 8, 100082
Jeandel, A and L. Garosi (2018), ‘Gait abnormalities in brachycephalic breeds: Should we be more concerned?’, Veterinary Record, 182 (6), 164–66
Julius K9. (2020a), ‘Dog harness safety misinformation’, available at https://julius-k9.co.uk/blog/dog-harness-safety-misinformation, accessed 18 March 2023
Julius K9. (2020b), ‘Scientific biomechanical study of Julius K9 harnesses’, available at https://julius-k9.co.uk/blog/idc-harness-biomechanical-study, accessed 23 March 2023
Kano, W., S. Rahal, F. Agostinho, L. Mesquita, R. Santos, F. Monteiro, M. Castillho and A. Melchert (2016), ‘Kinetic and temporospatial gait parameters in a heterogeneous group of dogs’, BMC Veterinary Research, 12 (2)
Ker, S., and A. Ramalingam (2013), ‘Is 30 the Magic Number? Issues in Sample Size Estimation’, National Journal of Community Medicine, 4 (1), 175–79
King, M and H. Zohny (2021), ‘Animal researchers shoulder a psychological burden that animal ethics committees ought to address’, Journal of Medical Ethics, 48 (5), 299–03
Kumpulainen, P., A. Cardó, S. Somppi, H. Törnqvist, H. Väätäjä, P. Majaranta, Y. Gizatdinova, C. Antink, V. Surakka, M. Kujala, O. Vainio and A. Vehkaoja (2021), ‘Dog behaviour classification with movement sensors placed on the harness and the collar’, Applied Animal Behaviour Science, 241, 105393
Lafuente, M., L. Provis and E. Schmalz (2019), ‘Effects of restrictive and non-restrictive harnesses on shoulder extension in dogs at walk and trot’, Veterinary Record, 184(2), 64
Landsberg, G., L. Radosta and L. Ackerman. (2023), Behavioural Problems of the Dog and Cat. (3rd edn.), Amsterdam, The Netherlands: Elsevier Health Sciences
Laverack, K., T. Pike, J. Cooper and E. Frasnelli (2021), ‘The effect of sex and age on paw use within a large sample of dogs (Canis familiaris)’, Applied Animal Behaviour Science, 238, 105298
Lorke, M., M. Willen, K. Lucas, M. Beyerbach, P. Wefstaedt, H. Murua Escobar and I. Nolte (2017), ‘Comparative kinematic gait analysis in young and old beagle dogs’, Journal of Veterinary Science, 18 (4), 521
Maliye, S and J. Marshall (2016), ‘Objective assessment of the compensatory effect of clinical hind limb lameness in horses: 37 cases (2011–2014)’, Journal of the American Veterinary Medical Association, 249 (8), 940–44
McIntyre, C. (2019), ‘Treadmills for canine conditioning programs: What you should know?’, available at http://www.mcrehabilitation.com/blog/treadmills-for-canine-conditioning-programs-what-you-should-know, accessed 30 March 2023
McMillan, T and K. Spaulding (2022), ‘Handling shelter dogs’, Animal Behavior for Shelter Veterinarians and Staff, 8, 177–04
Meeson, R., R. Todhunter, G. Blunn, G. Nuki and A. Pitsillides (2019), ‘Spontaneous dog osteoarthritis – a One Medicine vision’, Nature Reviews Rheumatology, 15, 273–87
Mocchi, M., S. Dotti, M. Bue, R. Villa, E. Bari, S. Perteghella, M. Torre and S. Grolli (2020), ‘Veterinary regenerative medicine for musculoskeletal disorders: Can mesenchymal stem/stromal cells and their secretome be the new frontier?’ Cells, 9 (6), 1453
Moro, M., G. Marchesi, F. Hesse, F. Odone and M. Casadio (2022), ‘Markerless vs. marker-based gait analysis: A proof of concept study’, Sensors (Basel), 22 (5), 2011
Morse, J. (2015), ‘Analytic strategies and sample size’, Qualitative Health Research, 25 (10), 1317–318
Murray, J., R. Kinsman, M. Lord, R. Da Costa, J. Woodward, S. Owczarczak-Garstecka, S. Tasker, T. Knowles and R. Casey (2021), ‘“Generation pup” – protocol for a longitudinal study of dog behaviour and health’, BMC Veterinary Research, 17 (1), 1
Nagymáté, G., R. Kiss and O. Biksi (2018), ‘Biomechanical analysis of the kinematics of different dog harnesses’, Julius K9 Research Report, available at https://julius-k9.com/pdf/biomechanical-analysis.pdf, accessed 17 March 2023
National Dog Warden Association. (2012), ‘Guidance for handling dogs: The use of equipment to ensure the health and safety of handlers, others and the welfare of dogs’, available at http://www.ndwa.co.uk/media/guidance-for-handling-dogs.pdf, accessed 21 March 2023
O’Neill, D., N. Coulson, D. Church and D. Brodbelt (2017), ‘Demography and disorders of German Shepherd Dogs under primary veterinary care in the UK’, Canine Genetics and Epidemiology, 4 (7)
Packer, R., A. Hendricks and C. Burn (2023), ‘Do dog owners perceive the clinical signs related to conformational inherited disorders as ‘normal’ for the breed? A potential constraint to improving canine welfare’, Animal Welfare, 21 (S1), 81–93
Packer, R. and M. Tivers (2015), ‘Strategies for the management and prevention of confirmation-related respiratory disorders in brachycephalic dogs’, Veterinary Medicine (Auckland, NZ), 6, 219–32
Pálya, Z., K. Rácz, G. Nagymáté and R. Kiss (2022), ‘Development of a detailed canine gait analysis method for evaluating harnesses: A pilot study’, PLOS ONE, 17 (3), e0264299
Peham, C., S. Limbeck, K. Galla and B. Bockstahler (2013), ‘Pressure distribution under three different types of harnesses used for guide dogs’, The Veterinary Journal, 18 (1), 93–98
Piccione, G., S. Casella., M. Panzera., C. Giannetto and F. Fazio (2012), ‘Effect of moderate treadmill exercise on some physiological parameters in untrained Beagle dogs’, Experimental Animals, 61 (5), 511–15
Preston, T., W. Baltzer and S. Trost (2012), ‘Accelerometer validity and placement for detection of changes in physical activity in dogs under controlled conditions on a treadmill’, Research in Veterinary Science, 93 (1), 412–16
Quintic Software (2009) Quintic Biomechanics Software, version 33, Birmingham: Quintic Consultancy Ltd
Rabe-Hesketh, S. and A. Skrondal (2010), ‘Generalized linear mixed models’, International Encyclopedia of Education (Third Edition), 171–77, Florida: CRC Press
Ramey, C., A. Mastali., C. Anderson., W. Stull., T. Starner and M. Jackson (2022), ‘WAG’D: Towards a wearable activity and gait detection monitor for sled dogs’, Proceedings of the Ninth International Conference on Animal-Computer Interaction, 18, 1–13
Rocznik, D., D. Sinn., S. Thomas and S. Gosling (2014), ‘Criterion analysis and content validity for standardized behavioral tests in a detector-dog breeding program’, Journal of Forensic Sciences, 60 (1), 213–21
Ruffwear. (2022), ‘Choosing the right dog harness’, available at https://ruffwear.com/pages/choosing-the-right-dog-harness, accessed 8 April 2023
Salonen, M., S. Mikkola, E. Hakanen., S. Sulkama., J. Puuruned and H. Lohi (2021), ‘Reliability and validity of a dog personality and unwanted behaviour study’, Animals, 11 (5), 1234
Sandberg, G. (2022), ‘An open-source MATLAB app for canine kinematics in three-dimensions’, ProQuest, 29063013
Schwencke, M., L. Smolders, N. Bergknut, P. Gustås, B. Meij and H. Hazewinkel (2012), ‘Soft tissue artifact in canine kinematic gait analysis’, Veterinary Surgery, 41(7), 829–37
Shih, H., M. Paterson, F. Georgiou, N. Pachana and C. Phillips (2020), ‘Who is pulling the leash? Effects of human gender and dog sex on human-dog dyads when walking on-leash’, Animals, 10 (10), 1894
Shih, H., C. Phillips, D. Mills, Y. Yang, F. Georgiou and M. Paterson (2021), ‘Dog pulling on the leash: Effects of restraint by a neck collar vs. a chest harness’, Frontiers in Veterinary Science, 8
Simmons, D., J. Benskin, J. Cosgrove, B. Duncker, D. Ekman, C. Martyniuk and J. Sherry (2015), ‘Omics for aquatic ecotoxicology: Control of extraneous variability to enhance the analysis of environmental effects’, Environmental Toxicology and Chemistry, 34 (8), 1693–704
Söhnel, K., M. Fischer and K. Häusler (2022), ‘Treadmill vs. overground trotting – a comparison of two kinetic measurement systems’, Research in Veterinary Science, 150, 149–55
Stroup, W. (2015). ‘Rethinking the analysis of non-normal data in plant and soil science’, Agronomy Journal, 107 (2), 811–27
Suresh, K. (2011), ‘An overview of randomization techniques: An unbiased assessment of outcome in clinical research’, Journal of Human Reproductive Sciences, 4 (1), 8–11
Takáčová, D., L. Skurková, L. Mesarčová, L. Lešková, L. Kottferová, A. Packová, D. Vajányi and J. Kottferová (2021), ‘Dog tethering in Slovakia: Legal, ethical and behavioral aspects and dog welfare implications’, Animals, 11 (3), 594
Tanneberger, A and C. Ciupitu-Plath (2017), ‘Nurses’ weight bias in caring for obese patients: Do weight controllability beliefs influence the provision of care to obese patients?’, Clinical Nursing Research, 27 (4), 414–32
Tasker, L., S. Getty, J. Briggs and V. Benka (2018), ‘Exploring the gaps in practical ethical guidance for animal welfare considerations of Field interventions and innovations targeting dogs and cats’, Animals, 8 (2), 19
Thompkins, A., G. Deshpande, P. Waggoner and J. Katz (2016), ‘Functional magnetic resonance imaging of the domestic dog: Research, methodology, and conceptual issues’, Comparative Cognition and Behavior Reviews, 11, 63–82
Thompson, M. (2020), ‘Harnesses: The good, the bad and the ugly’, available at https://www.happyhounduniversity.com/single-post/harnesses-the-good-the-bad-and-the-ugly, accessed 19 April 2023
Topál, J., V. Román and B. Turcsán (2019), ‘The dog ( Canis familiaris ) as a translational model of autism: It is high time we move from promise to reality’, WIREs Cognitive Science, 10 (4), 1495
Torres, B., Y. Fu, G. Sandberg and S. Budsberg (2017), ‘Pelvic limb kinematics in the dog with and without a stifle orthosis’, Veterinary Surgery, 46 (5), 642–52
Torres, B., N. Moëns, S. Al-Nadaf, L. Reynolds, Y. Fu and S. Budsberg (2013), ‘Comparison of overground and treadmill-based gaits of dogs’, American Journal of Veterinary Research, 74 (4), 535–41
Townsend, L., L. Dixon and L. Buckley (2022), ‘Lead pulling as a welfare concern in pet dogs: What can veterinary professionals learn from current research?’, Veterinary Record, 191 (10), 1627
UK General Data Protection Regulation 2021, c. 12, available at: https://www.legislation.gov.uk/ukpga/2018/12/contents/enacted, accessed 12 April 2023
Vetter, T. (2017), ‘Fundamentals of research data and variables: The devils is in the details’, Anesthesia and Analgesia, 125 (4), 1375–380
VioVet. (2023), ‘Halti’, available at https://www.viovet.co.uk/brands/816-Halti/page-1.html, accessed 23 March 2023
Virag, Y., M. Gumpenberger, A. Tichy, C. Lutonsky, C. Peham and B. Bockstahler (2022), ‘Center of pressure and ground reaction forces in Labrador and golden retrievers with and without hip dysplasia at 4, 8, and 12 months of age’, Frontiers in Veterinary Science, 9, 1087693
Voss, K., T. Wiestner., L. Galeandro., M. Hässig and P. Montavon (2011), ‘Effect of dog breed and body conformation on vertical ground reaction forces, impulses, and stance times’, Veterinary and Comparative Orthopaedics and Traumatology, 24 (2), 106–12
Wang, R., J. Xu, W. Zhang, J. Gao, Y. Li and F. Chen (2022), ‘Reliability analysis of complex electromechanical systems: State of the art, challenges, and prospects’, Quality and Reliability Engineering International, 38 (7), 3935–69
Ward, M. (2021), What Dogs Want: An Illustrated Guide for Happy Dog Care and Training, London: Bloomsbury Publishing
Wiener, P and M. Haskell (2016), ‘Use of questionnaire-based data to assess dog personality’, Journal of Veterinary Behavior, 16, 81–85
Winter, J. (2013), ‘Using the student's t-test with extremely small sample sizes’, Practical Assessment, Research, and Evaluation, 18 (10), https://doi.org/10.7275/e4r6-dj05
Woodin, J. (2015), ‘How to research Interculturally and ethically’, Research Methods in Intercultural Communication, 7, 103–19
Zeira, O., S. Scaccia, L. Pettinari, E. Ghezzi, N. Asiag, L. Martinelli, D. Zahirpour, M. Dumas, M. Konar, D. Lupi, L. Fiette, L. Pascucci, L. Leonardi, A. Cliff, G. Alessandri, A. Pessina, D. Spaziante and M. Aralla (2018), ‘Intra-articular administration of autologous micro-fragmented adipose tissue in dogs with spontaneous osteoarthritis: Safety, feasibility, and clinical outcomes’, Stem Cells Translational Medicine, 7 (11), 819–28
Zink, C. (2019), ‘To harness or not to harness? That is the question…’, Dogwise Publishing, available at http://www.highlineaussies.com/files/To-Harness-Or-Not-To-Harness-That-Is-The-Question-Avidog-Zink-Ventures.pdf, accessed 17 March 2023
Żuk, B and K. Księżopolska-Orłowska (2015), ‘Patterns of compensation of functional deficits of the knee joint in patients with juvenile idiopathic arthritis’, Reumatologia, 4, 192–99
Asymmetry: A lack of symmetry; differences in the behaviour of the limbs during movement.
Biomechanical: The study of mechanics and movement or structure of living organisms.
Brachycephalic breeds: dog breeds with flat faces and shortened snouts, which can lead to health problems.
Conformational disorders: Animals with exaggerated body shape, structure or appearance, which can negatively impact their health and welfare.
Gait: The pattern of steps of an animal at a certain speed or pace.
Gait analysis: the measurement of how the body moves during locomotion.
Generalised Linear Mixed Model (GLMM): a statistical regression model which is able to include randomeffects from different distributions.
Habituation: The decreased response to a certain stimulus after repeated exposure.
Kruskal Wallis: a nonparametric test which is rank-based which tests to see if there are statistical significant differences between two or more groups.
Longitudinal studies: A study that takes place over a long period of time; subjects are followed over time with either continuous or varied monitoring to see long-term impacts.
Markers: Reflective markers are placed on the subject; these are then tracked by cameras and used to create a 3D model of the subject’s movement.
Morphological: Refers to the size, shape and structure of the body.
Non-restrictive harness: A harness that does not restrict movement.
Peak vertical force: The maximum force exerted perpendicular to the surface during stance phase.
Post-hoc test: A test used to analyse the results of an experiment with more than two groups to identify where the differences lie between the groups.
Restrictive harness: A harness that restricts movement in some way.
Straight front harness: this type of dog harness has a single, usually thick strap that goes horizontally across the dog’s chest.
Tukey honest significant difference post-hoc: a test to see the significant differences between different groups.
Y-front harness: this type of dog harness has y-shaped straps that runs between the dog’s front legs over the shoulder.
To cite this paper please use the following details: Dowdeswell, L. and Churchill, L. (2024), 'The Influence of Harness Design on Forelimb Biomechanics in Pet Dogs', Reinvention: an International Journal of Undergraduate Research, Volume 17, Issue S1, https://reinventionjournal.org/article/view/1372. Date accessed [insert date]. If you cite this article or use it in any teaching or other related activities please let us know by e-mailing us at Reinventionjournal@warwick.ac.uk.